Fix resolving symbols for section-relative relocations

Also fixes MIPS `j` handling when jumping within the function.

Reworks `ObjReloc` struct to be a little more sensible.

.github/workflows/build.yaml

@@ -110,11 +110,6 @@ jobs:
             name: linux-aarch64
             build: zigbuild
             features: default
-          - platform: ubuntu-latest
-            target: armv7-unknown-linux-musleabi
-            name: linux-armv7l
-            build: zigbuild
-            features: default
           - platform: windows-latest
             target: i686-pc-windows-msvc
             name: windows-x86
@@ -147,7 +142,11 @@ jobs:
         uses: actions/checkout@v4
       - name: Install cargo-zigbuild
         if: matrix.build == 'zigbuild'
-        run: pip install ziglang==0.13.0 cargo-zigbuild==0.19.1
+        run: |
+          python3 -m venv .venv
+          . .venv/bin/activate
+          echo PATH=$PATH >> $GITHUB_ENV
+          pip install ziglang==0.13.0 cargo-zigbuild==0.19.1
       - name: Setup Rust toolchain
         uses: dtolnay/rust-toolchain@stable
         with:

Cargo.toml

@@ -13,7 +13,7 @@ strip = "debuginfo"
 codegen-units = 1
 
 [workspace.package]
-version = "2.0.0"
+version = "2.3.2"
 authors = ["Luke Street <luke@street.dev>"]
 edition = "2021"
 license = "MIT OR Apache-2.0"

config.schema.json

@@ -133,6 +133,13 @@
         },
         "metadata": {
           "ref": "#/$defs/metadata"
+        },
+        "symbol_mappings": {
+          "type": "object",
+          "description": "Manual symbol mappings from target to base.",
+          "additionalProperties": {
+            "type": "string"
+          }
         }
       }
     },

deny.toml

@@ -70,7 +70,6 @@ feature-depth = 1
 # A list of advisory IDs to ignore. Note that ignored advisories will still
 # output a note when they are encountered.
 ignore = [
-    "RUSTSEC-2024-0370",
     #{ id = "RUSTSEC-0000-0000", reason = "you can specify a reason the advisory is ignored" },
     #"a-crate-that-is-yanked@0.1.1", # you can also ignore yanked crate versions if you wish
     #{ crate = "a-crate-that-is-yanked@0.1.1", reason = "you can specify why you are ignoring the yanked crate" },
@@ -240,7 +239,7 @@ allow-git = []
 
 [sources.allow-org]
 # github.com organizations to allow git sources for
-github = ["encounter"]
+github = ["notify-rs"]
 # gitlab.com organizations to allow git sources for
 gitlab = []
 # bitbucket.org organizations to allow git sources for

objdiff-cli/Cargo.toml

@@ -11,7 +11,6 @@ description = """
 A local diffing tool for decompilation projects.
 """
 publish = false
-build = "build.rs"
 
 [dependencies]
 anyhow = "1.0"
@@ -29,3 +28,6 @@ supports-color = "3.0"
 time = { version = "0.3", features = ["formatting", "local-offset"] }
 tracing = "0.1"
 tracing-subscriber = { version = "0.3", features = ["env-filter"] }
+
+[target.'cfg(target_env = "musl")'.dependencies]
+mimalloc = "0.1"

objdiff-cli/build.rs

@@ -1,9 +0,0 @@
-fn main() {
-    let output = std::process::Command::new("git")
-        .args(["rev-parse", "HEAD"])
-        .output()
-        .expect("Failed to execute git");
-    let rev = String::from_utf8(output.stdout).expect("Failed to parse git output");
-    println!("cargo:rustc-env=GIT_COMMIT_SHA={rev}");
-    println!("cargo:rustc-rerun-if-changed=.git/HEAD");
-}

objdiff-cli/src/argp_version.rs

@@ -31,10 +31,9 @@ where T: FromArgs
             Ok(v) => {
                 if v.version {
                     println!(
-                        "{} {} {}",
+                        "{} {}",
                         command_name.first().unwrap_or(&""),
                         env!("CARGO_PKG_VERSION"),
-                        env!("GIT_COMMIT_SHA"),
                     );
                     std::process::exit(0);
                 } else {

objdiff-cli/src/cmd/diff.rs

@@ -102,26 +102,32 @@ pub fn run(args: Args) -> Result<()> {
                     let unit_path =
                         PathBuf::from_str(u).ok().and_then(|p| fs::canonicalize(p).ok());
 
-                    let Some(object) = project_config.objects.iter_mut().find_map(|obj| {
-                        if obj.name.as_deref() == Some(u) {
+                    let Some(object) = project_config
+                        .units
+                        .as_deref_mut()
+                        .unwrap_or_default()
+                        .iter_mut()
+                        .find_map(|obj| {
+                            if obj.name.as_deref() == Some(u) {
+                                resolve_paths(obj);
+                                return Some(obj);
+                            }
+
+                            let up = unit_path.as_deref()?;
+
                             resolve_paths(obj);
-                            return Some(obj);
-                        }
 
-                        let up = unit_path.as_deref()?;
+                            if [&obj.base_path, &obj.target_path]
+                                .into_iter()
+                                .filter_map(|p| p.as_ref().and_then(|p| p.canonicalize().ok()))
+                                .any(|p| p == up)
+                            {
+                                return Some(obj);
+                            }
 
-                        resolve_paths(obj);
-
-                        if [&obj.base_path, &obj.target_path]
-                            .into_iter()
-                            .filter_map(|p| p.as_ref().and_then(|p| p.canonicalize().ok()))
-                            .any(|p| p == up)
-                        {
-                            return Some(obj);
-                        }
-
-                        None
-                    }) else {
+                            None
+                        })
+                    else {
                         bail!("Unit not found: {}", u)
                     };
 
@@ -129,7 +135,13 @@ pub fn run(args: Args) -> Result<()> {
                 } else if let Some(symbol_name) = &args.symbol {
                     let mut idx = None;
                     let mut count = 0usize;
-                    for (i, obj) in project_config.objects.iter_mut().enumerate() {
+                    for (i, obj) in project_config
+                        .units
+                        .as_deref_mut()
+                        .unwrap_or_default()
+                        .iter_mut()
+                        .enumerate()
+                    {
                         resolve_paths(obj);
 
                         if obj
@@ -148,7 +160,7 @@ pub fn run(args: Args) -> Result<()> {
                     }
                     match (count, idx) {
                         (0, None) => bail!("Symbol not found: {}", symbol_name),
-                        (1, Some(i)) => &mut project_config.objects[i],
+                        (1, Some(i)) => &mut project_config.units_mut()[i],
                         (2.., Some(_)) => bail!(
                             "Multiple instances of {} were found, try specifying a unit",
                             symbol_name
@@ -303,7 +315,7 @@ fn find_function(obj: &ObjInfo, name: &str) -> Option<SymbolRef> {
     None
 }
 
-#[allow(dead_code)]
+#[expect(dead_code)]
 struct FunctionDiffUi {
     relax_reloc_diffs: bool,
     left_highlight: HighlightKind,
@@ -758,7 +770,7 @@ impl FunctionDiffUi {
         self.scroll_y += self.per_page / if half { 2 } else { 1 };
     }
 
-    #[allow(clippy::too_many_arguments)]
+    #[expect(clippy::too_many_arguments)]
     fn print_sym(
         &self,
         out: &mut Text<'static>,

objdiff-cli/src/cmd/report.rs

@@ -94,7 +94,7 @@ fn generate(args: GenerateArgs) -> Result<()> {
     };
     info!(
         "Generating report for {} units (using {} threads)",
-        project.objects.len(),
+        project.units().len(),
         if args.deduplicate { 1 } else { rayon::current_num_threads() }
     );
 
@@ -103,7 +103,7 @@ fn generate(args: GenerateArgs) -> Result<()> {
     let mut existing_functions: HashSet<String> = HashSet::new();
     if args.deduplicate {
         // If deduplicating, we need to run single-threaded
-        for object in &mut project.objects {
+        for object in project.units.as_deref_mut().unwrap_or_default() {
             if let Some(unit) = report_object(
                 object,
                 project_dir,
@@ -116,7 +116,9 @@ fn generate(args: GenerateArgs) -> Result<()> {
         }
     } else {
         let vec = project
-            .objects
+            .units
+            .as_deref_mut()
+            .unwrap_or_default()
             .par_iter_mut()
             .map(|object| {
                 report_object(
@@ -132,7 +134,7 @@ fn generate(args: GenerateArgs) -> Result<()> {
     }
     let measures = units.iter().flat_map(|u| u.measures.into_iter()).collect();
     let mut categories = Vec::new();
-    for category in &project.progress_categories {
+    for category in project.progress_categories() {
         categories.push(ReportCategory {
             id: category.id.clone(),
             name: category.name.clone(),
@@ -199,7 +201,7 @@ fn report_object(
             .unwrap_or_default(),
         auto_generated: object.metadata.as_ref().and_then(|m| m.auto_generated),
     };
-    let mut measures = Measures::default();
+    let mut measures = Measures { total_units: 1, ..Default::default() };
     let mut sections = vec![];
     let mut functions = vec![];
 
@@ -237,7 +239,7 @@ fn report_object(
         }
 
         for (symbol, symbol_diff) in section.symbols.iter().zip(&section_diff.symbols) {
-            if symbol.size == 0 {
+            if symbol.size == 0 || symbol.flags.0.contains(ObjSymbolFlags::Hidden) {
                 continue;
             }
             if let Some(existing_functions) = &mut existing_functions {
@@ -280,6 +282,7 @@ fn report_object(
     if metadata.complete.unwrap_or(false) {
         measures.complete_code = measures.total_code;
         measures.complete_data = measures.total_data;
+        measures.complete_units = 1;
     }
     measures.calc_fuzzy_match_percent();
     measures.calc_matched_percent();

objdiff-cli/src/main.rs

@@ -2,6 +2,12 @@ mod argp_version;
 mod cmd;
 mod util;
 
+// musl's allocator is very slow, so use mimalloc when targeting musl.
+// Otherwise, use the system allocator to avoid extra code size.
+#[cfg(target_env = "musl")]
+#[global_allocator]
+static ALLOC: mimalloc::MiMalloc = mimalloc::MiMalloc;
+
 use std::{env, ffi::OsStr, fmt::Display, path::PathBuf, str::FromStr};
 
 use anyhow::{Error, Result};

objdiff-core/Cargo.toml

@@ -17,8 +17,8 @@ crate-type = ["cdylib", "rlib"]
 
 [features]
 all = ["config", "dwarf", "mips", "ppc", "x86", "arm", "bindings"]
-any-arch = [] # Implicit, used to check if any arch is enabled
-config = ["globset", "semver", "serde_json", "serde_yaml"]
+any-arch = ["bimap"] # Implicit, used to check if any arch is enabled
+config = ["bimap", "globset", "semver", "serde_json", "serde_yaml"]
 dwarf = ["gimli"]
 mips = ["any-arch", "rabbitizer"]
 ppc = ["any-arch", "cwdemangle", "cwextab", "ppc750cl"]
@@ -27,8 +27,12 @@ arm = ["any-arch", "cpp_demangle", "unarm", "arm-attr"]
 bindings = ["serde_json", "prost", "pbjson"]
 wasm = ["bindings", "console_error_panic_hook", "console_log"]
 
+[package.metadata.docs.rs]
+features = ["all"]
+
 [dependencies]
 anyhow = "1.0"
+bimap = { version = "0.6", features = ["serde"], optional = true }
 byteorder = "1.5"
 filetime = "0.2"
 flagset = "0.4"
@@ -57,7 +61,7 @@ gimli = { version = "0.31", default-features = false, features = ["read-all"], o
 
 # ppc
 cwdemangle = { version = "1.0", optional = true }
-cwextab = { version = "0.2", optional = true }
+cwextab = { version = "1.0.2", optional = true }
 ppc750cl = { version = "0.3", optional = true }
 
 # mips

objdiff-core/protos/report.proto

@@ -32,6 +32,10 @@ message Measures {
   uint64 complete_data = 13;
   // Completed (or "linked") data percent
   float complete_data_percent = 14;
+  // Total number of units
+  uint32 total_units = 15;
+  // Completed (or "linked") units
+  uint32 complete_units = 16;
 }
 
 // Project progress report

objdiff-core/src/arch/mips.rs

@@ -83,7 +83,7 @@ impl ObjArch for ObjArchMips {
         &self,
         address: u64,
         code: &[u8],
-        _section_index: usize,
+        section_index: usize,
         relocations: &[ObjReloc],
         line_info: &BTreeMap<u64, u32>,
         config: &DiffObjConfig,
@@ -140,11 +140,18 @@ impl ObjArch for ObjArchMips {
                     | OperandType::cpu_label
                     | OperandType::cpu_branch_target_label => {
                         if let Some(reloc) = reloc {
-                            if matches!(&reloc.target_section, Some(s) if s == ".text")
-                                && reloc.target.address > start_address
-                                && reloc.target.address < end_address
+                            // If the relocation target is within the current function, we can
+                            // convert it into a relative branch target. Note that we check
+                            // target_address > start_address instead of >= so that recursive
+                            // tail calls are not considered branch targets.
+                            let target_address =
+                                reloc.target.address.checked_add_signed(reloc.addend);
+                            if reloc.target.orig_section_index == Some(section_index)
+                                && matches!(target_address, Some(addr) if addr > start_address && addr < end_address)
                             {
-                                args.push(ObjInsArg::BranchDest(reloc.target.address));
+                                let target_address = target_address.unwrap();
+                                args.push(ObjInsArg::BranchDest(target_address));
+                                branch_dest = Some(target_address);
                             } else {
                                 push_reloc(&mut args, reloc)?;
                                 branch_dest = None;

objdiff-core/src/arch/mod.rs

@@ -1,11 +1,13 @@
-use std::{borrow::Cow, collections::BTreeMap};
+use std::{borrow::Cow, collections::BTreeMap, ffi::CStr};
 
 use anyhow::{bail, Result};
+use byteorder::ByteOrder;
 use object::{Architecture, File, Object, ObjectSymbol, Relocation, RelocationFlags, Symbol};
 
 use crate::{
     diff::DiffObjConfig,
     obj::{ObjIns, ObjReloc, ObjSection},
+    util::ReallySigned,
 };
 
 #[cfg(feature = "arm")]
@@ -17,6 +19,97 @@ pub mod ppc;
 #[cfg(feature = "x86")]
 pub mod x86;
 
+/// Represents the type of data associated with an instruction
+pub enum DataType {
+    Int8,
+    Int16,
+    Int32,
+    Int64,
+    Int128,
+    Float,
+    Double,
+    Bytes,
+    String,
+}
+
+impl DataType {
+    pub fn display_bytes<Endian: ByteOrder>(&self, bytes: &[u8]) -> Option<String> {
+        if self.required_len().is_some_and(|l| bytes.len() < l) {
+            return None;
+        }
+
+        match self {
+            DataType::Int8 => {
+                let i = i8::from_ne_bytes(bytes.try_into().unwrap());
+                if i < 0 {
+                    format!("Int8: {:#x} ({:#x})", i, ReallySigned(i))
+                } else {
+                    format!("Int8: {:#x}", i)
+                }
+            }
+            DataType::Int16 => {
+                let i = Endian::read_i16(bytes);
+                if i < 0 {
+                    format!("Int16: {:#x} ({:#x})", i, ReallySigned(i))
+                } else {
+                    format!("Int16: {:#x}", i)
+                }
+            }
+            DataType::Int32 => {
+                let i = Endian::read_i32(bytes);
+                if i < 0 {
+                    format!("Int32: {:#x} ({:#x})", i, ReallySigned(i))
+                } else {
+                    format!("Int32: {:#x}", i)
+                }
+            }
+            DataType::Int64 => {
+                let i = Endian::read_i64(bytes);
+                if i < 0 {
+                    format!("Int64: {:#x} ({:#x})", i, ReallySigned(i))
+                } else {
+                    format!("Int64: {:#x}", i)
+                }
+            }
+            DataType::Int128 => {
+                let i = Endian::read_i128(bytes);
+                if i < 0 {
+                    format!("Int128: {:#x} ({:#x})", i, ReallySigned(i))
+                } else {
+                    format!("Int128: {:#x}", i)
+                }
+            }
+            DataType::Float => {
+                format!("Float: {}", Endian::read_f32(bytes))
+            }
+            DataType::Double => {
+                format!("Double: {}", Endian::read_f64(bytes))
+            }
+            DataType::Bytes => {
+                format!("Bytes: {:#?}", bytes)
+            }
+            DataType::String => {
+                format!("String: {:?}", CStr::from_bytes_until_nul(bytes).ok()?)
+            }
+        }
+        .into()
+    }
+
+    fn required_len(&self) -> Option<usize> {
+        match self {
+            DataType::Int8 => Some(1),
+            DataType::Int16 => Some(2),
+            DataType::Int32 => Some(4),
+            DataType::Int64 => Some(8),
+            DataType::Int128 => Some(16),
+            DataType::Float => Some(4),
+            DataType::Double => Some(8),
+            DataType::Bytes => None,
+            DataType::String => None,
+        }
+    }
+}
+
 pub trait ObjArch: Send + Sync {
     fn process_code(
         &self,
@@ -42,6 +135,12 @@ pub trait ObjArch: Send + Sync {
 
     fn symbol_address(&self, symbol: &Symbol) -> u64 { symbol.address() }
 
+    fn guess_data_type(&self, _instruction: &ObjIns) -> Option<DataType> { None }
+
+    fn display_data_type(&self, _ty: DataType, bytes: &[u8]) -> Option<String> {
+        Some(format!("Bytes: {:#x?}", bytes))
+    }
+
     // Downcast methods
     #[cfg(feature = "ppc")]
     fn ppc(&self) -> Option<&ppc::ObjArchPpc> { None }

objdiff-core/src/arch/ppc.rs

@@ -1,15 +1,16 @@
 use std::{borrow::Cow, collections::BTreeMap};
 
 use anyhow::{bail, ensure, Result};
+use byteorder::BigEndian;
 use cwextab::{decode_extab, ExceptionTableData};
 use object::{
     elf, File, Object, ObjectSection, ObjectSymbol, Relocation, RelocationFlags, RelocationTarget,
     Symbol, SymbolKind,
 };
-use ppc750cl::{Argument, InsIter, GPR};
+use ppc750cl::{Argument, InsIter, Opcode, GPR};
 
 use crate::{
-    arch::{ObjArch, ProcessCodeResult},
+    arch::{DataType, ObjArch, ProcessCodeResult},
     diff::DiffObjConfig,
     obj::{ObjIns, ObjInsArg, ObjInsArgValue, ObjReloc, ObjSection, ObjSymbol},
 };
@@ -186,6 +187,34 @@ impl ObjArch for ObjArchPpc {
         }
     }
 
+    fn guess_data_type(&self, instruction: &ObjIns) -> Option<super::DataType> {
+        // Always shows the first string of the table. Not ideal, but it's really hard to find
+        // the actual string being referenced.
+        if instruction.reloc.as_ref().is_some_and(|r| r.target.name.starts_with("@stringBase")) {
+            return Some(DataType::String);
+        }
+
+        match Opcode::from(instruction.op as u8) {
+            Opcode::Lbz | Opcode::Lbzu | Opcode::Lbzux | Opcode::Lbzx => Some(DataType::Int8),
+            Opcode::Lhz | Opcode::Lhzu | Opcode::Lhzux | Opcode::Lhzx => Some(DataType::Int16),
+            Opcode::Lha | Opcode::Lhau | Opcode::Lhaux | Opcode::Lhax => Some(DataType::Int16),
+            Opcode::Lwz | Opcode::Lwzu | Opcode::Lwzux | Opcode::Lwzx => Some(DataType::Int32),
+            Opcode::Lfs | Opcode::Lfsu | Opcode::Lfsux | Opcode::Lfsx => Some(DataType::Float),
+            Opcode::Lfd | Opcode::Lfdu | Opcode::Lfdux | Opcode::Lfdx => Some(DataType::Double),
+
+            Opcode::Stb | Opcode::Stbu | Opcode::Stbux | Opcode::Stbx => Some(DataType::Int8),
+            Opcode::Sth | Opcode::Sthu | Opcode::Sthux | Opcode::Sthx => Some(DataType::Int16),
+            Opcode::Stw | Opcode::Stwu | Opcode::Stwux | Opcode::Stwx => Some(DataType::Int32),
+            Opcode::Stfs | Opcode::Stfsu | Opcode::Stfsux | Opcode::Stfsx => Some(DataType::Float),
+            Opcode::Stfd | Opcode::Stfdu | Opcode::Stfdux | Opcode::Stfdx => Some(DataType::Double),
+            _ => None,
+        }
+    }
+
+    fn display_data_type(&self, ty: DataType, bytes: &[u8]) -> Option<String> {
+        ty.display_bytes::<BigEndian>(bytes)
+    }
+
     fn ppc(&self) -> Option<&ObjArchPpc> { Some(self) }
 }
 
@@ -303,9 +332,13 @@ fn decode_exception_info(file: &File<'_>) -> Result<Option<BTreeMap<usize, Excep
             continue;
         };
         let data = match decode_extab(extab_data) {
-            Some(decoded_data) => decoded_data,
-            None => {
-                log::warn!("Exception table decoding failed for function {}", extab_func_name);
+            Ok(decoded_data) => decoded_data,
+            Err(e) => {
+                log::warn!(
+                    "Exception table decoding failed for function {}, reason: {}",
+                    extab_func_name,
+                    e.to_string()
+                );
                 return Ok(None);
             }
         };

objdiff-core/src/bindings/diff.rs

@@ -70,9 +70,13 @@ impl FunctionDiff {
         //     let (_section, symbol) = object.section_symbol(symbol_ref);
         //     Symbol::from(symbol)
         // });
-        let instructions = symbol_diff.instructions.iter().map(InstructionDiff::from).collect();
+        let instructions = symbol_diff
+            .instructions
+            .iter()
+            .map(|ins_diff| InstructionDiff::new(object, ins_diff))
+            .collect();
         Self {
-            symbol: Some(Symbol::from(symbol)),
+            symbol: Some(Symbol::new(symbol)),
             // diff_symbol,
             instructions,
             match_percent: symbol_diff.match_percent,
@@ -90,8 +94,8 @@ impl DataDiff {
     }
 }
 
-impl<'a> From<&'a ObjSymbol> for Symbol {
-    fn from(value: &'a ObjSymbol) -> Self {
+impl Symbol {
+    pub fn new(value: &ObjSymbol) -> Self {
         Self {
             name: value.name.to_string(),
             demangled_name: value.demangled_name.clone(),
@@ -122,29 +126,29 @@ fn symbol_flags(value: ObjSymbolFlagSet) -> u32 {
     flags
 }
 
-impl<'a> From<&'a ObjIns> for Instruction {
-    fn from(value: &'a ObjIns) -> Self {
+impl Instruction {
+    pub fn new(object: &ObjInfo, instruction: &ObjIns) -> Self {
         Self {
-            address: value.address,
-            size: value.size as u32,
-            opcode: value.op as u32,
-            mnemonic: value.mnemonic.clone(),
-            formatted: value.formatted.clone(),
-            arguments: value.args.iter().map(Argument::from).collect(),
-            relocation: value.reloc.as_ref().map(Relocation::from),
-            branch_dest: value.branch_dest,
-            line_number: value.line,
-            original: value.orig.clone(),
+            address: instruction.address,
+            size: instruction.size as u32,
+            opcode: instruction.op as u32,
+            mnemonic: instruction.mnemonic.clone(),
+            formatted: instruction.formatted.clone(),
+            arguments: instruction.args.iter().map(Argument::new).collect(),
+            relocation: instruction.reloc.as_ref().map(|reloc| Relocation::new(object, reloc)),
+            branch_dest: instruction.branch_dest,
+            line_number: instruction.line,
+            original: instruction.orig.clone(),
         }
     }
 }
 
-impl<'a> From<&'a ObjInsArg> for Argument {
-    fn from(value: &'a ObjInsArg) -> Self {
+impl Argument {
+    pub fn new(value: &ObjInsArg) -> Self {
         Self {
             value: Some(match value {
                 ObjInsArg::PlainText(s) => argument::Value::PlainText(s.to_string()),
-                ObjInsArg::Arg(v) => argument::Value::Argument(ArgumentValue::from(v)),
+                ObjInsArg::Arg(v) => argument::Value::Argument(ArgumentValue::new(v)),
                 ObjInsArg::Reloc => argument::Value::Relocation(ArgumentRelocation {}),
                 ObjInsArg::BranchDest(dest) => argument::Value::BranchDest(*dest),
             }),
@@ -152,8 +156,8 @@ impl<'a> From<&'a ObjInsArg> for Argument {
     }
 }
 
-impl From<&ObjInsArgValue> for ArgumentValue {
-    fn from(value: &ObjInsArgValue) -> Self {
+impl ArgumentValue {
+    pub fn new(value: &ObjInsArgValue) -> Self {
         Self {
             value: Some(match value {
                 ObjInsArgValue::Signed(v) => argument_value::Value::Signed(*v),
@@ -164,42 +168,39 @@ impl From<&ObjInsArgValue> for ArgumentValue {
     }
 }
 
-impl<'a> From<&'a ObjReloc> for Relocation {
-    fn from(value: &ObjReloc) -> Self {
+impl Relocation {
+    pub fn new(object: &ObjInfo, reloc: &ObjReloc) -> Self {
         Self {
-            r#type: match value.flags {
+            r#type: match reloc.flags {
                 object::RelocationFlags::Elf { r_type } => r_type,
                 object::RelocationFlags::MachO { r_type, .. } => r_type as u32,
                 object::RelocationFlags::Coff { typ } => typ as u32,
                 object::RelocationFlags::Xcoff { r_rtype, .. } => r_rtype as u32,
                 _ => unreachable!(),
             },
-            type_name: String::new(), // TODO
-            target: Some(RelocationTarget::from(&value.target)),
+            type_name: object.arch.display_reloc(reloc.flags).into_owned(),
+            target: Some(RelocationTarget {
+                symbol: Some(Symbol::new(&reloc.target)),
+                addend: reloc.addend,
+            }),
         }
     }
 }
 
-impl<'a> From<&'a ObjSymbol> for RelocationTarget {
-    fn from(value: &'a ObjSymbol) -> Self {
-        Self { symbol: Some(Symbol::from(value)), addend: value.addend }
-    }
-}
-
-impl<'a> From<&'a ObjInsDiff> for InstructionDiff {
-    fn from(value: &'a ObjInsDiff) -> Self {
+impl InstructionDiff {
+    pub fn new(object: &ObjInfo, instruction_diff: &ObjInsDiff) -> Self {
         Self {
-            instruction: value.ins.as_ref().map(Instruction::from),
-            diff_kind: DiffKind::from(value.kind) as i32,
-            branch_from: value.branch_from.as_ref().map(InstructionBranchFrom::from),
-            branch_to: value.branch_to.as_ref().map(InstructionBranchTo::from),
-            arg_diff: value.arg_diff.iter().map(ArgumentDiff::from).collect(),
+            instruction: instruction_diff.ins.as_ref().map(|ins| Instruction::new(object, ins)),
+            diff_kind: DiffKind::from(instruction_diff.kind) as i32,
+            branch_from: instruction_diff.branch_from.as_ref().map(InstructionBranchFrom::new),
+            branch_to: instruction_diff.branch_to.as_ref().map(InstructionBranchTo::new),
+            arg_diff: instruction_diff.arg_diff.iter().map(ArgumentDiff::new).collect(),
         }
     }
 }
 
-impl From<&Option<ObjInsArgDiff>> for ArgumentDiff {
-    fn from(value: &Option<ObjInsArgDiff>) -> Self {
+impl ArgumentDiff {
+    pub fn new(value: &Option<ObjInsArgDiff>) -> Self {
         Self { diff_index: value.as_ref().map(|v| v.idx as u32) }
     }
 }
@@ -228,8 +229,8 @@ impl From<ObjDataDiffKind> for DiffKind {
     }
 }
 
-impl<'a> From<&'a ObjInsBranchFrom> for InstructionBranchFrom {
-    fn from(value: &'a ObjInsBranchFrom) -> Self {
+impl InstructionBranchFrom {
+    pub fn new(value: &ObjInsBranchFrom) -> Self {
         Self {
             instruction_index: value.ins_idx.iter().map(|&x| x as u32).collect(),
             branch_index: value.branch_idx as u32,
@@ -237,8 +238,8 @@ impl<'a> From<&'a ObjInsBranchFrom> for InstructionBranchFrom {
     }
 }
 
-impl<'a> From<&'a ObjInsBranchTo> for InstructionBranchTo {
-    fn from(value: &'a ObjInsBranchTo) -> Self {
+impl InstructionBranchTo {
+    pub fn new(value: &ObjInsBranchTo) -> Self {
         Self { instruction_index: value.ins_idx as u32, branch_index: value.branch_idx as u32 }
     }
 }

objdiff-core/src/bindings/report.rs

@@ -8,9 +8,10 @@ use serde_json::error::Category;
 include!(concat!(env!("OUT_DIR"), "/objdiff.report.rs"));
 include!(concat!(env!("OUT_DIR"), "/objdiff.report.serde.rs"));
 
-pub const REPORT_VERSION: u32 = 1;
+pub const REPORT_VERSION: u32 = 2;
 
 impl Report {
+    /// Attempts to parse the report as binary protobuf or JSON.
     pub fn parse(data: &[u8]) -> Result<Self> {
         if data.is_empty() {
             bail!(std::io::Error::from(std::io::ErrorKind::UnexpectedEof));
@@ -25,6 +26,7 @@ impl Report {
         Ok(report)
     }
 
+    /// Attempts to parse the report as JSON, migrating from the legacy report format if necessary.
     fn from_json(bytes: &[u8]) -> Result<Self, serde_json::Error> {
         match serde_json::from_slice::<Self>(bytes) {
             Ok(report) => Ok(report),
@@ -43,16 +45,23 @@ impl Report {
         }
     }
 
+    /// Migrates the report to the latest version.
+    /// Fails if the report version is newer than supported.
     pub fn migrate(&mut self) -> Result<()> {
         if self.version == 0 {
             self.migrate_v0()?;
         }
+        if self.version == 1 {
+            self.migrate_v1()?;
+        }
         if self.version != REPORT_VERSION {
             bail!("Unsupported report version: {}", self.version);
         }
         Ok(())
     }
 
+    /// Adds `complete_code`, `complete_data`, `complete_code_percent`, and `complete_data_percent`
+    /// to measures, and sets `progress_categories` in unit metadata.
     fn migrate_v0(&mut self) -> Result<()> {
         let Some(measures) = &mut self.measures else {
             bail!("Missing measures in report");
@@ -61,15 +70,16 @@ impl Report {
             let Some(unit_measures) = &mut unit.measures else {
                 bail!("Missing measures in report unit");
             };
-            let Some(metadata) = &mut unit.metadata else {
-                bail!("Missing metadata in report unit");
+            let mut complete = false;
+            if let Some(metadata) = &mut unit.metadata {
+                if metadata.module_name.is_some() || metadata.module_id.is_some() {
+                    metadata.progress_categories = vec!["modules".to_string()];
+                } else {
+                    metadata.progress_categories = vec!["dol".to_string()];
+                }
+                complete = metadata.complete.unwrap_or(false);
             };
-            if metadata.module_name.is_some() || metadata.module_id.is_some() {
-                metadata.progress_categories = vec!["modules".to_string()];
-            } else {
-                metadata.progress_categories = vec!["dol".to_string()];
-            }
-            if metadata.complete.unwrap_or(false) {
+            if complete {
                 unit_measures.complete_code = unit_measures.total_code;
                 unit_measures.complete_data = unit_measures.total_data;
                 unit_measures.complete_code_percent = 100.0;
@@ -84,10 +94,42 @@ impl Report {
             measures.complete_data += unit_measures.complete_data;
         }
         measures.calc_matched_percent();
+        self.calculate_progress_categories();
         self.version = 1;
         Ok(())
     }
 
+    /// Adds `total_units` and `complete_units` to measures.
+    fn migrate_v1(&mut self) -> Result<()> {
+        let Some(total_measures) = &mut self.measures else {
+            bail!("Missing measures in report");
+        };
+        for unit in &mut self.units {
+            let Some(measures) = &mut unit.measures else {
+                bail!("Missing measures in report unit");
+            };
+            let complete = unit.metadata.as_ref().and_then(|m| m.complete).unwrap_or(false) as u32;
+            let progress_categories =
+                unit.metadata.as_ref().map(|m| m.progress_categories.as_slice()).unwrap_or(&[]);
+            measures.total_units = 1;
+            measures.complete_units = complete;
+            total_measures.total_units += 1;
+            total_measures.complete_units += complete;
+            for id in progress_categories {
+                if let Some(category) = self.categories.iter_mut().find(|c| &c.id == id) {
+                    let Some(measures) = &mut category.measures else {
+                        bail!("Missing measures in category");
+                    };
+                    measures.total_units += 1;
+                    measures.complete_units += complete;
+                }
+            }
+        }
+        self.version = 2;
+        Ok(())
+    }
+
+    /// Calculate progress categories based on unit metadata.
     pub fn calculate_progress_categories(&mut self) {
         for unit in &self.units {
             let Some(metadata) = unit.metadata.as_ref() else {
@@ -117,6 +159,72 @@ impl Report {
             measures.calc_matched_percent();
         }
     }
+
+    /// Split the report into multiple reports based on progress categories.
+    /// Assumes progress categories are in the format `version`, `version.category`.
+    /// This is a hack for projects that generate all versions in a single report.
+    pub fn split(self) -> Vec<(String, Report)> {
+        let mut reports = Vec::new();
+        // Map units to Option to allow taking ownership
+        let mut units = self.units.into_iter().map(Some).collect::<Vec<_>>();
+        for category in &self.categories {
+            if category.id.contains(".") {
+                // Skip subcategories
+                continue;
+            }
+            fn is_sub_category(id: &str, parent: &str, sep: char) -> bool {
+                id.starts_with(parent)
+                    && id.get(parent.len()..).map_or(false, |s| s.starts_with(sep))
+            }
+            let mut sub_categories = self
+                .categories
+                .iter()
+                .filter(|c| is_sub_category(&c.id, &category.id, '.'))
+                .cloned()
+                .collect::<Vec<_>>();
+            // Remove category prefix
+            for sub_category in &mut sub_categories {
+                sub_category.id = sub_category.id[category.id.len() + 1..].to_string();
+            }
+            let mut sub_units = units
+                .iter_mut()
+                .filter_map(|opt| {
+                    let unit = opt.as_mut()?;
+                    let metadata = unit.metadata.as_ref()?;
+                    if metadata.progress_categories.contains(&category.id) {
+                        opt.take()
+                    } else {
+                        None
+                    }
+                })
+                .collect::<Vec<_>>();
+            for sub_unit in &mut sub_units {
+                // Remove leading version/ from unit name
+                if let Some(name) =
+                    sub_unit.name.strip_prefix(&category.id).and_then(|s| s.strip_prefix('/'))
+                {
+                    sub_unit.name = name.to_string();
+                }
+                // Filter progress categories
+                let Some(metadata) = sub_unit.metadata.as_mut() else {
+                    continue;
+                };
+                metadata.progress_categories = metadata
+                    .progress_categories
+                    .iter()
+                    .filter(|c| is_sub_category(c, &category.id, '.'))
+                    .map(|c| c[category.id.len() + 1..].to_string())
+                    .collect();
+            }
+            reports.push((category.id.clone(), Report {
+                measures: category.measures,
+                units: sub_units,
+                version: self.version,
+                categories: sub_categories,
+            }));
+        }
+        reports
+    }
 }
 
 impl Measures {
@@ -176,6 +284,8 @@ impl AddAssign for Measures {
         self.matched_functions += other.matched_functions;
         self.complete_code += other.complete_code;
         self.complete_data += other.complete_data;
+        self.total_units += other.total_units;
+        self.complete_units += other.complete_units;
     }
 }
 

objdiff-core/src/config/mod.rs

@@ -1,77 +1,100 @@
 use std::{
+    fs,
     fs::File,
-    io::Read,
+    io::{BufReader, BufWriter, Read},
     path::{Path, PathBuf},
 };
 
 use anyhow::{anyhow, Context, Result};
+use bimap::BiBTreeMap;
 use filetime::FileTime;
 use globset::{Glob, GlobSet, GlobSetBuilder};
 
-#[inline]
-fn bool_true() -> bool { true }
-
-#[derive(Default, Clone, serde::Deserialize)]
+#[derive(Default, Clone, serde::Serialize, serde::Deserialize)]
 pub struct ProjectConfig {
-    #[serde(default)]
+    #[serde(default, skip_serializing_if = "Option::is_none")]
     pub min_version: Option<String>,
-    #[serde(default)]
+    #[serde(default, skip_serializing_if = "Option::is_none")]
     pub custom_make: Option<String>,
-    #[serde(default)]
+    #[serde(default, skip_serializing_if = "Option::is_none")]
     pub custom_args: Option<Vec<String>>,
-    #[serde(default)]
+    #[serde(default, skip_serializing_if = "Option::is_none")]
     pub target_dir: Option<PathBuf>,
-    #[serde(default)]
+    #[serde(default, skip_serializing_if = "Option::is_none")]
     pub base_dir: Option<PathBuf>,
-    #[serde(default = "bool_true")]
-    pub build_base: bool,
-    #[serde(default)]
-    pub build_target: bool,
-    #[serde(default)]
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub build_base: Option<bool>,
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub build_target: Option<bool>,
+    #[serde(default, skip_serializing_if = "Option::is_none")]
     pub watch_patterns: Option<Vec<Glob>>,
-    #[serde(default, alias = "units")]
-    pub objects: Vec<ProjectObject>,
-    #[serde(default)]
-    pub progress_categories: Vec<ProjectProgressCategory>,
+    #[serde(default, alias = "objects", skip_serializing_if = "Option::is_none")]
+    pub units: Option<Vec<ProjectObject>>,
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub progress_categories: Option<Vec<ProjectProgressCategory>>,
 }
 
-#[derive(Default, Clone, serde::Deserialize)]
+impl ProjectConfig {
+    #[inline]
+    pub fn units(&self) -> &[ProjectObject] { self.units.as_deref().unwrap_or_default() }
+
+    #[inline]
+    pub fn units_mut(&mut self) -> &mut Vec<ProjectObject> {
+        self.units.get_or_insert_with(Vec::new)
+    }
+
+    #[inline]
+    pub fn progress_categories(&self) -> &[ProjectProgressCategory] {
+        self.progress_categories.as_deref().unwrap_or_default()
+    }
+
+    #[inline]
+    pub fn progress_categories_mut(&mut self) -> &mut Vec<ProjectProgressCategory> {
+        self.progress_categories.get_or_insert_with(Vec::new)
+    }
+}
+
+#[derive(Default, Clone, serde::Serialize, serde::Deserialize)]
 pub struct ProjectObject {
-    #[serde(default)]
+    #[serde(default, skip_serializing_if = "Option::is_none")]
     pub name: Option<String>,
-    #[serde(default)]
+    #[serde(default, skip_serializing_if = "Option::is_none")]
     pub path: Option<PathBuf>,
-    #[serde(default)]
+    #[serde(default, skip_serializing_if = "Option::is_none")]
     pub target_path: Option<PathBuf>,
-    #[serde(default)]
+    #[serde(default, skip_serializing_if = "Option::is_none")]
     pub base_path: Option<PathBuf>,
-    #[serde(default)]
+    #[serde(default, skip_serializing_if = "Option::is_none")]
     #[deprecated(note = "Use metadata.reverse_fn_order")]
     pub reverse_fn_order: Option<bool>,
-    #[serde(default)]
+    #[serde(default, skip_serializing_if = "Option::is_none")]
     #[deprecated(note = "Use metadata.complete")]
     pub complete: Option<bool>,
-    #[serde(default)]
+    #[serde(default, skip_serializing_if = "Option::is_none")]
     pub scratch: Option<ScratchConfig>,
-    #[serde(default)]
+    #[serde(default, skip_serializing_if = "Option::is_none")]
     pub metadata: Option<ProjectObjectMetadata>,
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub symbol_mappings: Option<SymbolMappings>,
 }
 
-#[derive(Default, Clone, serde::Deserialize)]
+pub type SymbolMappings = BiBTreeMap<String, String>;
+
+#[derive(Default, Clone, serde::Serialize, serde::Deserialize)]
 pub struct ProjectObjectMetadata {
-    #[serde(default)]
+    #[serde(default, skip_serializing_if = "Option::is_none")]
     pub complete: Option<bool>,
-    #[serde(default)]
+    #[serde(default, skip_serializing_if = "Option::is_none")]
     pub reverse_fn_order: Option<bool>,
-    #[serde(default)]
+    #[serde(default, skip_serializing_if = "Option::is_none")]
     pub source_path: Option<String>,
-    #[serde(default)]
+    #[serde(default, skip_serializing_if = "Option::is_none")]
     pub progress_categories: Option<Vec<String>>,
-    #[serde(default)]
+    #[serde(default, skip_serializing_if = "Option::is_none")]
     pub auto_generated: Option<bool>,
 }
 
-#[derive(Default, Clone, serde::Deserialize)]
+#[derive(Default, Clone, serde::Serialize, serde::Deserialize)]
 pub struct ProjectProgressCategory {
     #[serde(default)]
     pub id: String,
@@ -112,32 +135,36 @@ impl ProjectObject {
     }
 
     pub fn complete(&self) -> Option<bool> {
-        #[allow(deprecated)]
+        #[expect(deprecated)]
         self.metadata.as_ref().and_then(|m| m.complete).or(self.complete)
     }
 
     pub fn reverse_fn_order(&self) -> Option<bool> {
-        #[allow(deprecated)]
+        #[expect(deprecated)]
         self.metadata.as_ref().and_then(|m| m.reverse_fn_order).or(self.reverse_fn_order)
     }
 
     pub fn hidden(&self) -> bool {
         self.metadata.as_ref().and_then(|m| m.auto_generated).unwrap_or(false)
     }
+
+    pub fn source_path(&self) -> Option<&String> {
+        self.metadata.as_ref().and_then(|m| m.source_path.as_ref())
+    }
 }
 
 #[derive(Default, Clone, Eq, PartialEq, serde::Deserialize, serde::Serialize)]
 pub struct ScratchConfig {
-    #[serde(default)]
+    #[serde(default, skip_serializing_if = "Option::is_none")]
     pub platform: Option<String>,
-    #[serde(default)]
+    #[serde(default, skip_serializing_if = "Option::is_none")]
     pub compiler: Option<String>,
-    #[serde(default)]
+    #[serde(default, skip_serializing_if = "Option::is_none")]
     pub c_flags: Option<String>,
-    #[serde(default)]
+    #[serde(default, skip_serializing_if = "Option::is_none")]
     pub ctx_path: Option<PathBuf>,
-    #[serde(default)]
-    pub build_ctx: bool,
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub build_ctx: Option<bool>,
 }
 
 pub const CONFIG_FILENAMES: [&str; 3] = ["objdiff.json", "objdiff.yml", "objdiff.yaml"];
@@ -150,13 +177,13 @@ pub const DEFAULT_WATCH_PATTERNS: &[&str] = &[
 #[derive(Clone, Eq, PartialEq)]
 pub struct ProjectConfigInfo {
     pub path: PathBuf,
-    pub timestamp: FileTime,
+    pub timestamp: Option<FileTime>,
 }
 
 pub fn try_project_config(dir: &Path) -> Option<(Result<ProjectConfig>, ProjectConfigInfo)> {
     for filename in CONFIG_FILENAMES.iter() {
         let config_path = dir.join(filename);
-        let Ok(mut file) = File::open(&config_path) else {
+        let Ok(file) = File::open(&config_path) else {
             continue;
         };
         let metadata = file.metadata();
@@ -165,9 +192,10 @@ pub fn try_project_config(dir: &Path) -> Option<(Result<ProjectConfig>, ProjectC
                 continue;
             }
             let ts = FileTime::from_last_modification_time(&metadata);
+            let mut reader = BufReader::new(file);
             let mut result = match filename.contains("json") {
-                true => read_json_config(&mut file),
-                false => read_yml_config(&mut file),
+                true => read_json_config(&mut reader),
+                false => read_yml_config(&mut reader),
             };
             if let Ok(config) = &result {
                 // Validate min_version if present
@@ -175,12 +203,41 @@ pub fn try_project_config(dir: &Path) -> Option<(Result<ProjectConfig>, ProjectC
                     result = Err(e);
                 }
             }
-            return Some((result, ProjectConfigInfo { path: config_path, timestamp: ts }));
+            return Some((result, ProjectConfigInfo { path: config_path, timestamp: Some(ts) }));
         }
     }
     None
 }
 
+pub fn save_project_config(
+    config: &ProjectConfig,
+    info: &ProjectConfigInfo,
+) -> Result<ProjectConfigInfo> {
+    if let Some(last_ts) = info.timestamp {
+        // Check if the file has changed since we last read it
+        if let Ok(metadata) = fs::metadata(&info.path) {
+            let ts = FileTime::from_last_modification_time(&metadata);
+            if ts != last_ts {
+                return Err(anyhow!("Config file has changed since last read"));
+            }
+        }
+    }
+    let mut writer =
+        BufWriter::new(File::create(&info.path).context("Failed to create config file")?);
+    let ext = info.path.extension().and_then(|ext| ext.to_str()).unwrap_or("json");
+    match ext {
+        "json" => serde_json::to_writer_pretty(&mut writer, config).context("Failed to write JSON"),
+        "yml" | "yaml" => {
+            serde_yaml::to_writer(&mut writer, config).context("Failed to write YAML")
+        }
+        _ => Err(anyhow!("Unknown config file extension: {ext}")),
+    }?;
+    let file = writer.into_inner().context("Failed to flush file")?;
+    let metadata = file.metadata().context("Failed to get file metadata")?;
+    let ts = FileTime::from_last_modification_time(&metadata);
+    Ok(ProjectConfigInfo { path: info.path.clone(), timestamp: Some(ts) })
+}
+
 fn validate_min_version(config: &ProjectConfig) -> Result<()> {
     let Some(min_version) = &config.min_version else { return Ok(()) };
     let version = semver::Version::parse(env!("CARGO_PKG_VERSION"))